# Sample clip attribution The demo's sample clips are short excerpts of openly licensed, public-domain or freely redistributable video, redistributed here under their terms; each excerpt carries the licence or terms of its source. The demo is non-commercial and no fee is charged for any of these files. Every clip ships in two sizes, `-360p.mp4` (640×360, read at the demo's 384×216 and 640×360 settings) and `-720p.mp4` (1280×720, read at 1280×720); the 1.9:1 Netflix clips are 640×338 and 1280×676. Changes made to every clip: cut to the segment below, audio removed, scaled with a Lanczos filter, and encoded with ffmpeg/libx264 (`-preset veryslow`, the CRF below, `aq-mode=3`, keyframe every 2 s) as H.264 High, yuv420p, BT.709, MP4 with faststart, for browser playback. H.264 was kept rather than AV1 because Safari plays AV1 only on devices with an AV1 hardware decoder. ## Blender Foundation open movies (CC BY 3.0) Licence: Creative Commons Attribution 3.0, . The Tears of Steel excerpts were also centre-cropped from 2.4:1 (1920×800) to 16:9 (1422×800) before scaling. All three were **retimed to 30 fps (1.25x)**: every 24 fps frame plays once at 30 fps, timestamps only, no re-encoding (`tools/web/retime-clips.py`). | Files | Work | Author | Source | Segment | CRF 360p / 720p | |---|---|---|---|---|---| | `tos-canal-*.mp4` | *Tears of Steel* (2012) | (CC) Blender Foundation, mango.blender.org | | 4:18.0–4:30.0 (12 s, 24 fps) | 22 / 24 | | `tos-bridge-*.mp4` | *Tears of Steel* (2012) | (CC) Blender Foundation, mango.blender.org | | 0:26.0–0:39.0 (13 s, 24 fps) | 22 / 24 | | `bbb-tree-*.mp4` | *Big Buck Bunny* (2008) | (c) copyright 2008, Blender Foundation, peach.blender.org | | 2:07.0–2:21.0 (14 s, 24 fps) | 22 / 24 | The Tears of Steel soundtrack is under a different licence (CC BY-ND 3.0); no audio is included. ## SVT Open Content Video Test Suite 2022 – Natural Complexity (CC BY 4.0) *SVT Open Content Video Test Suite 2022 – Natural Complexity* by Sveriges Television AB (SVT), Video Core team, is licensed under Creative Commons Attribution 4.0 International, . - Suite page and licence statement: - Documentation (v1.3), whose "Intentions, license and restrictions of use" section states the CC BY 4.0 licence: - Masters: `ftp://svtopencontent.svt.se/pub/svt_videotestsuite_natural_complexity/REC709/` (SDR Rec. 709 mezzanines, 3840×2160p50, JPEG 2000 lossless, YUV 4:4:4 12-bit) Additional changes: 3840×2160 downscaled to 1280×720 and 640×360, every second frame dropped (50 → 25 fps), 4:4:4 12-bit converted to 4:2:0 8-bit, then **retimed to 30 fps (1.2x)**: every frame plays once at 30 fps (timestamps only, `tools/web/retime-clips.py`). | Files | Sequence | Master file | Segment | CRF 360p / 720p | |---|---|---|---|---| | `svt-waterfall-*.mp4` | waterfall (SDR) | `waterfall_JPEG2000_SDR_3840x2160p50_YUV444_12bit_Lossless.mov` | 0:10.0–0:20.0 (10 s) | 25 / 28 | | `svt-midnight-sun-*.mp4` | midnight_sun (SDR) | `midnight_sun_JPEG2000_SDR_3840x2160p50_YUV444_12bit_Lossless.mov` | 0:10.0–0:20.0 (10 s) | 25 / 28 | | `svt-smoke-sauna-*.mp4` | smoke_sauna (SDR) | `smoke_sauna_JPEG2000_SDR_3840x2160p50_YUV444_12bit_Lossless.mov` | 0:10.0–0:20.0 (10 s) | 25 / 28 | ## Xiph.org derf collection: JCT-VC / Vidyo conferencing sequences (public domain) The Xiph.org Video Test Media page (derf's collection), , labels each of these sequences "16:9 | 600 frames | public domain" (checked 2026-11-10; the page gives no per-file notes for them). They were originally contributed by Vidyo to the JCT-VC HEVC test set (class E, videoconferencing). Source files: lossless 1280×720p60 YUV 4:2:0 8-bit y4m, `https://media.xiph.org/video/derf/y4m/`. Additional changes: every second frame dropped (60 → 30 fps); the 720p file is not rescaled. | Files | Sequence | Source file | Segment | CRF 360p / 720p | |---|---|---|---|---| | `xiph-johnny-*.mp4` | Johnny | `Johnny_1280x720_60.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `xiph-kristen-sara-*.mp4` | KristenAndSara | `KristenAndSara_1280x720_60.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `xiph-four-people-*.mp4` | FourPeople | `FourPeople_1280x720_60.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `xiph-vidyo1-*.mp4` | vidyo1 | `vidyo1_720p_60fps.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `xiph-vidyo3-*.mp4` | vidyo3 | `vidyo3_720p_60fps.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `xiph-vidyo4-*.mp4` | vidyo4 | `vidyo4_720p_60fps.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | ## SVT High Definition Multi Format Test Set (2006) These excerpts come from *The SVT High Definition Multi Format Test Set*, Version 1.0, Feb. 2006, by Sveriges Television AB (SVT), taken from the SVT programme 'Fairytale'. They are distributed here under the set's own terms, not an open licence. This project uses them for developing, testing and presenting implementations of video coding standards (H.264, HEVC, AV1, VP9, AV2, VVC), and no fee is charged. The set's documentation, which its terms require to accompany any distributed part, ships next to the clips: [`SVT_MultiFormat_v10.pdf`](SVT_MultiFormat_v10.pdf) (from ). Terms, verbatim from that document (§1 and §2): > All sequences from this set may be distributed freely (complete or in parts) as long as the > copyright and the restrictions of the use of the sequences are not violated. > This document has to be distributed with any part of the test set that is distributed. > It is not allowed to charge a fee for the distribution of the sequences. Anybody receiving > material from this data set is kindly asked to further distribute these sequences to other parties > to reach a super distribution effect. > > **2. Copyright and Restrictions of Use** > > Individuals and organizations extracting sequences from this archive agree that the sequences > and all intellectual property rights therein remain the property of Sveriges Television AB > (SVT), Sweden. These sequences may only be used for the purpose of developing, testing and > presenting technology standards. SVT makes no warranties with respect to the materials and > expressly disclaim any warranties regarding their fitness for any purpose. Source files: the Xiph.org derf collection's y4m conversions, `https://media.xiph.org/video/derf/y4m/` (originally `ftp://vqeg.its.bldrdoc.gov/HDTV/SVT_MultiFormat/`). Additional changes: every second frame dropped (50 → 25 fps), then **retimed to 30 fps (1.2x)** (timestamps only, `tools/web/retime-clips.py`); crowd_run downscaled from 1080p (the others are SVT's own 720p versions and are not rescaled at 720p). | Files | Sequence | Source file | Segment | CRF 360p / 720p | |---|---|---|---|---| | `svt-old-town-cross-*.mp4` | OldTownCross | `old_town_cross_420_720p50.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `svt-in-to-tree-*.mp4` | IntoTree | `in_to_tree_420_720p50.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 22 / 24 | | `svt-crowd-run-*.mp4` | CrowdRun | `crowd_run_1080p50.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 25 / 28 | | `svt-park-joy-*.mp4` | ParkJoy | `park_joy_420_720p50.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 25 / 28 | | `svt-ducks-take-off-*.mp4` | DucksTakeOff | `ducks_take_off_420_720p50.y4m` | 0:00.0–0:10.0 (10 s, the whole sequence) | 25 / 28 | ## Netflix Open Content: El Fuente and Chimera (CC BY-NC-ND 4.0) Licence: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International, . The demo is non-commercial and no fee is charged. Each file is a format conversion of an unmodified, contiguous excerpt (10 s, 600 consecutive frames, or the whole sequence where it is shorter; every frame kept at the source's frame timing, full picture with no cropping): 10-bit converted to 8-bit, downscaled with a Lanczos filter from 4096×2160 to 1280×676 (`-720p.mp4`) and 640×338 (`-360p.mp4`) so the 1.9:1 frame is kept whole (the demo letterboxes it), and encoded to H.264 as described above (keyframe every 120 frames). Nothing else is changed. Source files: the Xiph.org derf collection, `https://media.xiph.org/video/derf//Netflix__4096x2160_60fps_10bit_420.y4m`; only the needed frames were fetched. | Files | Sequence | Source folder | Source frames | CRF 360p / 720p | |---|---|---|---|---| | `netflix-narrator-*.mp4` | Narrator (*El Fuente*) | `ElFuente/` | 0–299 (all 300, 5.0 s) | 22 / 24 | | `netflix-dinner-scene-*.mp4` | DinnerScene (*Chimera*) | `Chimera/` | 60–659 (10 s) | 22 / 24 | | `netflix-bar-scene-*.mp4` | BarScene (*Chimera*) | `Chimera/` | 60–659 (10 s) | 22 / 24 | | `netflix-crosswalk-*.mp4` | Crosswalk (*El Fuente*) | `ElFuente/` | 0–299 (all 300, 5.0 s) | 25 / 28 | | `netflix-tango-*.mp4` | Tango (*El Fuente*) | `ElFuente/` | 0–293 (all 294, 4.9 s) | 22 / 24 | | `netflix-boxing-practice-*.mp4` | BoxingPractice (*El Fuente*) | `ElFuente/` | 0–253 (all 254, 4.2 s) | 22 / 24 | | `netflix-dancers-*.mp4` | Dancers (*Chimera*) | `Chimera/` | 60–659 (10 s) | 22 / 24 | | `netflix-toddler-fountain-*.mp4` | ToddlerFountain (*Chimera*) | `Chimera/` | 60–659 (10 s) | 25 / 28 | Copyright notices, verbatim from each sequence's `Netflix__Copyright.txt` on the derf page: `netflix-narrator-*.mp4` (from ): ```text Digital Video Sequence "Narrator" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 300 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-dinner-scene-*.mp4` (from ): ```text Digital Video Sequence "Dinner Scene" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 1199 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-bar-scene-*.mp4` (from ): ```text Digital Video Sequence "Bar Scene" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 1199 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-crosswalk-*.mp4` (from ): ```text Digital Video Sequence "Crosswalk" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 300 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-tango-*.mp4` (from ): ```text Digital Video Sequence "Tango" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 294 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-boxing-practice-*.mp4` (from ): ```text Digital Video Sequence "Boxing Practice" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 254 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-dancers-*.mp4` (from ): ```text Digital Video Sequence "Dancers" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 1199 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` `netflix-toddler-fountain-*.mp4` (from ): ```text Digital Video Sequence "Toddler Fountain" Copyright (C) NETFLIX INC. 2015 4096x2160p, 10-bit, YCbCr 4:2:0, BT.709, 59.94 fps, 1199 frames COPYRIGHT AND LICENSE INFORMATION December, 2015 NETFLIX INC. 100 Winchester Circle, Los Gatos, CA 95032, USA The video sequence provided above and all intellectual property rights therein remain the property of Netflix Inc. This video sequence is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ ``` ## MOTChallenge MOT17 / MOT20 (CC BY-NC-SA 3.0) MOTChallenge MOT17 (Milan et al.) / MOT20 (Dendorfer et al.), Creative Commons Attribution-NonCommercial-ShareAlike 3.0, , . Source: the original JPEG frames of the training sequences. Changes: a 10 s excerpt (300 frames), scaled with Lanczos (full picture kept: MOT20-05's 1654×1080 becomes 1102×720 and 552×360 and is letterboxed by the demo), full-range BT.601 converted to limited-range BT.709, encoded as described above (`tools/web/make-mot-clips.py`), every frame played once at 30 fps (MOT20-05, a 25 fps sequence, is **retimed to 30 fps (1.2x)**). **These excerpts, and the pre-encoded streams made from them, are offered under the same CC BY-NC-SA 3.0 licence (ShareAlike), for non-commercial use only.** | Files | Sequence | Source frames | CRF 360p / 720p | |---|---|---|---| | `mot17-mall-walk-*.mp4` | MOT17-11 (moving camera, shopping mall, 1920×1080, 30 fps) | 601–900 (20.0–30.0 s) | 25 / 28 | | `mot17-night-crowd-*.mp4` | MOT17-04 (static elevated camera, night street, 1920×1080, 30 fps) | 211–510 (7.0–17.0 s) | 25 / 28 | | `mot20-dense-crowd-*.mp4` | MOT20-05 (~200 people, night, 1654×1080, 25 fps) | 2301–2600 (92.0–104.0 s) | 25 / 28 | The excerpts are the windows with the most frame-to-frame change times spatial detail (`make-mot-clips.py --pick`). ## Pre-encoded reference streams (`web/streams-ref/`) The demo's pre-encoded streams are further adaptations of the Xiph (public domain), Blender (CC BY 3.0), SVT 2022 (CC BY 4.0) and MOTChallenge (CC BY-NC-SA 3.0; the streams too are CC BY-NC-SA 3.0, and each carries that notice in the manifest and the demo's panel) excerpts above, under the same terms and credits: each is the hosted excerpt re-encoded by a third-party encoder (x264, openh264, x265, SVT-AV1, libaom, libvpx or NVENC; every command line is in `web/streams-ref/manifest.json` and `docs/results/reference-streams.md`), coded from the decoded frames of the hosted file, with for some streams scaling (854×480, 384×216), letterboxing and frame-rate sampling (every 2nd or 3rd frame: 15, 10 fps). No SVT Multi Format sequence is used. **No Netflix sequence has pre-encoded streams** (CC BY-NC-ND 4.0: no derivatives). The FoodMarket excerpt and its streams were removed on 2026-11-15. Only material whose terms allow redistribution is shipped here; the project's main evaluation sets (`/mldata`) do not. Johnny and FourPeople are also in the small `adv12` screening set (AOM CTC copies); being public domain, they may be redistributed.